Big Tech's $725B AI spend faces reckoning after chip selloff
TL;DR
- The Nasdaq 100 fell 4.1% last week and the Philadelphia semiconductor index sank 10%, its worst week since April 2025.
- Alphabet, Microsoft, Amazon and Meta have forecast up to $725 billion in AI-driven capex this calendar year.
- Wall Street analysts compiled by Bloomberg expect that figure to climb to nearly $900 billion in 2027.
Seven hundred and twenty-five billion dollars is a lot of capex to defend, and last week the market started asking the question out loud. Bloomberg reports that after a sharp wipeout in chips and the broader tech complex, pressure is building on the biggest AI spenders to justify their expenditures to traders with, in Bloomberg's phrase, increasingly itchy fingers hovering over their sell buttons.
The numbers underneath that framing are the part worth staring at. Information technology was the worst-performing group in the S&P 500 last week. The index itself slid 1.6%, the Nasdaq 100 lost 4.1%, and the Philadelphia Stock Exchange Semiconductor Index sank 10% for its worst week since April 2025. Against that backdrop, Alphabet, Microsoft, Amazon and Meta have forecast as much as $725 billion in capital expenditures this calendar year, with the analyst average compiled by Bloomberg pointing to nearly $900 billion in 2027.
The proximate spark, according to other reporting, was a Chinese model release. Fortune wrote that Moonshot AI's Kimi K3, pitched as competitive with the strongest closed frontier systems at a fraction of the cost, gave the market a fresh version of the DeepSeek shock: if capable models can be trained and served for much less, the revenue side of the hyperscaler capex equation looks less certain than it did a quarter ago.
The honest caveat is that the Bloomberg piece is a market read, not an earnings post-mortem. It does not break out how much of that $725 billion is contractually committed versus guidance the companies could quietly throttle, and it does not put current AI revenue next to the capex to show the gap. Take the specifics as reported, not settled, and watch the next set of hyperscaler prints for the harder answer.
What is worth watching from here is which of the four names uses the coming earnings season to show revenue actually pacing with the spend. The ones that can will get their multiples back. The ones that cannot are going to spend the rest of 2026 explaining themselves, and every open-weight release out of China will make that conversation louder.
Originally reported by bloomberg.com
Read the original article →Original headline: Bloomberg: Big Tech Needs to Justify AI Spending as Investors Dump Stocks, Nasdaq 100 Sheds 4.1% and Chips Fall 10% in Worst Week Since April